megatron

Install requirements.txt

Model can be downloaded from:-https://drive.google.com/drive/folders/1neHfilNkrCwtry3dA7bTCADH3EheKdYo?usp=sharing

This is a question answering system which is used to service users at websites.

Description

• It is quite often that we hire some people as customer service so they can interact with user's when they come on the website but what if we can automate this task?
• Because users usually come can ask similar questions and have similar requests a majority of this work can be automated.
• We create Megatron a question answering system.
• It can:
• • take input from user in text, audio or video format
• • classify user requests and provide the respective service or information.
• • If the user is not satified then can let someone attend the user.

Aim / Goal

• To automate customer service.

Skillset

• Python programming.
• NLP
• Data Validation
• Data preprocessing
• feature engineering
• model creation
• logging
• model fine tuning
• model selection

What i Learnt

• Industry standard OOP's based approach
• logging.
• clustering methodology for better model building.
• data science lifecycle
• The various nlp architectures like bert, GPT, albert, distilbert
• deployment on aws.

Special thanks to ineuron for teaching and walking me through this real life end to end project.